Fix drawing terminator, round usec

workaround for rounding us/iter
for some reason rounding is giving me wrong results not sure what i'm doing wrong :D just show 6 digits :P ok this gets the us/iter working, and it is more stable but the elapsed time still needs to be added
2025-01-08 18:34:46 -08:00 · 2025-01-05 14:29:27 -08:00 · 2025-01-05 14:05:24 -08:00 · 2025-01-04 18:46:51 -08:00 · 2025-01-04 14:25:25 -08:00 · 2025-01-04 12:13:27 -08:00
3 changed files with 212 additions and 140 deletions
--- a/mandel.s
+++ b/mandel.s
@ -1,16 +1,16 @@
 ; Our zero-page vars
-ox              = $80 ; fixed8.24: center point x
-oy              = $84 ; fixed8.24: center point y
-cx              = $88 ; fixed8.24: c_x
-cy              = $8c ; fixed8.24: c_y
+ox              = $80 ; fixed6.26: center point x
+oy              = $84 ; fixed6.26: center point y
+cx              = $88 ; fixed6.26: c_x
+cy              = $8c ; fixed6.26: c_y

-zx              = $90 ; fixed8.24: z_x
-zy              = $94 ; fixed8.24: z_y
-zx_2            = $98 ; fixed8.24: z_x^2
-zy_2            = $9c ; fixed8.24: z_y^2
+zx              = $90 ; fixed6.26: z_x
+zy              = $94 ; fixed6.26: z_y
+zx_2            = $98 ; fixed6.26: z_x^2
+zy_2            = $9c ; fixed6.26: z_y^2

-zx_zy           = $a0 ; fixed8.24: z_x * z_y
-dist            = $a4 ; fixed8.24: z_x^2 + z_y^2
+zx_zy           = $a0 ; fixed6.26: z_x * z_y
+dist            = $a4 ; fixed6.26: z_x^2 + z_y^2
 sx              = $a8 ; i16: screen pixel x
 sy              = $aa ; i16: screen pixel y
 z_buffer_active = $ac ; boolean: 1 if we triggered the lake, 0 if not
@ -31,10 +31,10 @@ chroma_offset   = $bb ; u8
 palette_ticks   = $bc ; u8
 chroma_ticks    = $bd ; u8
 count_frames    = $be ; u8
-count_pixels    = $bf ; u8
+; free space $bf

-total_pixels    = $c0 ; float48
-total_ms        = $c6 ; float48
+count_iters     = $c0 ; u16
+; free space c2-cb
 temp            = $cc ; u16
 temp2           = $ce ; u16

@ -59,10 +59,12 @@ LBUFF  = $0580 ; result buffer for FASC routine
 ; FP ROM routine vectors
 FASC   = $D8E6 ; FLOATING POINT TO ASCII (output in INBUFF, last char has high bit set)
 IFP    = $D9AA ; INTEGER TO FLOATING POINT CONVERSION (FR0:u16 -> FR0:float48)
+FPI    = $D9D2 ; floating point to integer
 FADD   = $DA66 ; ADDITION       (FR0 += FR1)
 FSUB   = $DA60 ; SUBTRACTION    (FR0 -= FR1)
 FMUL   = $DADB ; MULTIPLICATION (FR0 *= FR1)
 FDIV   = $DB28 ; DIVISION       (FR0 /= FR1)
+ZFR0   = $DA44 ; clear FR0
 ZF1    = $DA46 ; CLEAR ZERO PAGE FLOATING POINT NUMBER (XX)
 FLD0R  = $DD89 ; LOAD FR0 WITH FLOATING POINT NUMBER (YYXX)
 FLD1R  = $DD98 ; LOAD FR1 WITH FLOATING POINT NUMBER (YYXX)
@ -141,7 +143,7 @@ str_self:
    .byte "MANDEL-6502"
 str_self_end:
 str_speed:
-    .byte " ms/px"
+    .byte "us/iter: "
 str_speed_end:
 str_run:
    .byte " RUN"
@ -189,20 +191,38 @@ aspect:
    ;
    ; 184h is the equiv of 220.8h at square pixels
    ; 320 / 220.8 = 1.45 display aspect ratio
-aspect_x: ; fixed4.16 5/4
-    .word 5 << (12 - 2)
+aspect_x: ; fixed3.13 5/4
+    .word 5 << (13 - 2)

-aspect_y: ; fixed4.16 3/4
-    .word 3 << (12 - 2)
+aspect_y: ; fixed3.13 3/4
+    .word 3 << (13 - 2)

-ms_per_frame: ; float48 16.66666667
-    .byte 64  ; exponent/sign
-    .byte $16 ; BCD digits
+sec_per_frame: ; float48 00 . 01 66 66 66 67
+    .byte 63  ; exponent/sign - -1 bytes
+    .byte $01 ; BCD digits
    .byte $66
    .byte $66
    .byte $66
    .byte $67

+us_per_sec: ; float48 1e9 01 00 0,0 00 . 00
+    .byte 67  ; exponent/sign +3 bytes
+    .byte $01 ; BCD digits
+    .byte $00
+    .byte $00
+    .byte $00
+    .byte $00
+
+total_iters: ; float48
+    .repeat 6
+        .byte 0
+    .endrepeat
+
+total_sec: ; float48
+    .repeat 6
+        .byte 0
+    .endrepeat
+
 display_list_start:
    ; 24 lines overscan
    .repeat 3
@ -234,9 +254,9 @@ display_list_len = display_list_end - display_list_start
 color_map:
    .byte 0
    .repeat 85
-        .byte 1
-        .byte 2
-        .byte 3
+        .byte %01010101
+        .byte %10101010
+        .byte %11111111
    .endrepeat


@ -285,23 +305,34 @@ fill_masks:
    .byte %00000001
    .byte %00000000

+pixel_masks:
+    .byte %11111111
+    .byte %11110000
+    .byte %11000000
+
 viewport_zoom:
-    .byte 1
-    .byte 6
-    .byte 8
-    .byte 6
+    .byte 0
+    .byte 5
+    .byte 7
+    .byte 5
+    .byte 7
+    .byte 7

 viewport_ox:
-    .dword $00000000
-    .dword $ff110000
-    .dword $ff110000
-    .dword $fe400000
+    .dword ($00000000 & $3fffffff) << 2
+    .dword ($ff110000 & $3fffffff) << 2
+    .dword ($ff110000 & $3fffffff) << 2
+    .dword ($fe400000 & $3fffffff) << 2
+    .dword ($fe3b0000 & $3fffffff) << 2
+    .dword $fd220000

 viewport_oy:
-    .dword $00000000
-    .dword $ffb60000
-    .dword $ffbe0000
-    .dword $00000000
+    .dword ($00000000 & $3fffffff) << 2
+    .dword ($ffb60000 & $3fffffff) << 2
+    .dword ($ffbe0000 & $3fffffff) << 2
+    .dword ($00000000 & $3fffffff) << 2
+    .dword ($fffe0000 & $3fffffff) << 2
+    .dword $ff000000

 ; 2 + 9 * byte cycles
 .macro add bytes, dest, arg1, arg2
@ -453,20 +484,6 @@ viewport_oy:
    sta dest + 1
 .endmacro

-; input: arg as u8
-; input/output: dest as u16
-; clobbers a, x
-.macro sqr8_add16 dest, arg
-    ldx arg
-    clc
-    lda sqr_lobyte,x
-    adc dest
-    sta dest
-    lda sqr_hibyte,x
-    adc dest + 1
-    sta dest + 1
-.endmacro
-
 .segment "TABLES"
 ; lookup table for top byte -> PORTB value for bank-switch
 .align 256
@ -749,9 +766,8 @@ inner_loop:
    ; h1*h2*256*256 + h1*l2*256 + h2*l1*256 + l1*l2

    imul8 result, arg1, arg2, xe
-    lda #0
-    sta result + 2
-    sta result + 3
+
+    imul8 result + 2, arg1 + 1, arg2 + 1, xe

    imul8 inter, arg1 + 1, arg2, xe
    add16 result + 1, result + 1, inter
@ -761,9 +777,6 @@ inner_loop:
    add16 result + 1, result + 1, inter
    add_carry result + 3

-    imul8 inter, arg1 + 1, arg2 + 1, xe
-    add16 result + 2, result + 2, inter
-
    ; In case of negative inputs, adjust high word
    ; https://stackoverflow.com/a/28827013
    lda arg1 + 1
@ -796,9 +809,8 @@ arg2_pos:
        ; h*h*256*256 + h*l*256 + h*l*256 + l*l

        sqr8 result, arg
-        lda #0
-        sta result + 2
-        sta result + 3
+
+        sqr8 result + 2, arg + 1

        imul8 inter, arg + 1, arg, xe
        add16 result + 1, result + 1, inter
@ -806,8 +818,6 @@ arg2_pos:
        add16 result + 1, result + 1, inter
        add_carry result + 3

-        sqr8_add16 result + 2, arg + 1
-
        rts ; 6 cyc
    .endscope
 .endmacro
@ -875,8 +885,8 @@ next:

 .proc mandelbrot
    ; input:
-    ; cx: position scaled to 8.24 fixed point - -128..+127.9
-    ; cy: position scaled to 8.24
+    ; cx: position scaled to 6.26 fixed point - -32..+31.9
+    ; cy: position scaled to 6.26
    ;
    ; output:
    ; iter: iteration count at escape or 0
@ -927,6 +937,11 @@ next:
    sta z_buffer_end

 loop:
+    inc count_iters
+    bne low_iters
+    inc count_iters + 1
+low_iters:
+
    ; iter++ & max-iters break
    inc iter
    bne keep_going
@ -934,7 +949,7 @@ loop:
 keep_going:

    .macro quick_exit arg, max
-        ; arg: fixed8.24
+        ; arg: fixed6.26
        ; max: integer
        .local positive
        .local negative
@ -947,12 +962,12 @@ keep_going:
        bmi negative

    positive:
-        cmp #max
+        cmp #(max << 2)
        bmi all_done ; 'less than'
        jmp exit_path

    negative:
-        cmp #(256 - max)
+        cmp #(256 - (max << 2))
        beq first_equal ; 'equal' on first byte
        bpl all_done    ; 'greater than'

@ -972,7 +987,7 @@ keep_going:
    all_done:
    .endmacro

-    ; 8.24: (-128 .. 127.9)
+    ; 6.26: (-32 .. 31.9)
    ; zx = zx_2  - zy_2  + cx
    sub32 zx, zx_2, zy_2
    add32 zx, zx, cx
@ -983,9 +998,9 @@ keep_going:
    add32 zy, zy, cy
    quick_exit zy, 2

-    ; convert 8.24 -> 4.12: (-8 .. +7.9)
-    shift_round_16 zx, 4
-    shift_round_16 zy, 4
+    ; convert 6.26 -> 3.13: (-4 .. +3.9)
+    shift_round_16 zx, 3
+    shift_round_16 zy, 3

    ; zx_2 = zx * zx
    sqr16 zx_2, zx + 2
@ -1107,9 +1122,9 @@ enough:
 .endmacro

 .macro zoom_factor dest, src, aspect
-    ; output: dest: fixed8.24
-    ; input: src: fixed4.12
-    ; aspect: fixed4.12
+    ; output: dest: fixed6.26
+    ; input: src: fixed3.13
+    ; aspect: fixed3.13
    ; clobbers A, X, flags, etc
    copy16 dest, src
    scale_zoom dest
@ -1127,8 +1142,11 @@ enough:
    ; iter -> color
    ldx iter
    lda color_map,x
+    ldx fill_level
+    and pixel_masks,x
    sta pixel_color
-    lda #(255 - 3)
+    lda pixel_masks,x
+    eor #$ff
    sta pixel_mask

    ; sy -> line base address in temp
@ -1177,22 +1195,23 @@ point:
    ; pixel_mask <<= pixel_shift (shifting in ones)
    and #3
    sta pixel_shift
-    lda #3
-    sec
-    sbc pixel_shift
    tax
 shift_loop:
    beq shift_done
-    asl pixel_color
-    asl pixel_color
+    lsr pixel_color
+    lsr pixel_color
    sec
-    rol pixel_mask
+    ror pixel_mask
    sec
-    rol pixel_mask
+    ror pixel_mask
    dex
    jmp shift_loop
 shift_done:

+    ldy fill_level
+    ldx fill_masks,y
+    inx
+
    ; pixel_offset = temp >> 2
    lda temp
    lsr a
@ -1200,12 +1219,25 @@ shift_done:
    sta pixel_offset
    tay

+draw_pixel:
    ; read, mask, or, write
    lda (pixel_ptr),y
    and pixel_mask
    ora pixel_color
    sta (pixel_ptr),y

+    dex
+    beq done
+    clc
+    lda #40
+    adc pixel_ptr
+    sta pixel_ptr
+    lda #0
+    adc pixel_ptr + 1
+    sta pixel_ptr + 1
+    jmp draw_pixel
+
+done:
    rts
 .endproc

@ -1213,6 +1245,7 @@ shift_done:
    ; clobbers A, X
    .local loop
    .local done
+    .local padding
    ldx #0
 loop:
    cpx #len
@ -1220,11 +1253,27 @@ loop:
    txa
    tay
    lda (strptr),y
+    pha ; save the char for terminator check
+    and #$7f ; strip the high bit (terminator)
    tay
    lda char_map,y
    sta textbuffer + col,x
    inx
+
+    pla
+    bmi padding
    jmp loop
+
+padding:
+    ldy #32 ; space
+    lda char_map,y
+
+    cpx #len
+    beq done
+    sta textbuffer + col,x
+    inx
+    jmp padding
+
 done:
 .endmacro

@ -1367,7 +1416,7 @@ skip_char:

 plus:
    lda zoom
-    cmp #8
+    cmp #7
    bpl skip_char
    inc zoom
    jmp done
@ -1399,6 +1448,10 @@ number_keys:
    beq three
    cpy #KEY_4
    beq four
+    cpy #KEY_5
+    beq five
+    cpy #KEY_6
+    beq six
    jmp skip_char

 one:
@ -1412,6 +1465,12 @@ three:
    jmp load_key_viewport
 four:
    ldx #3
+    jmp load_key_viewport
+five:
+    ldx #4
+    jmp load_key_viewport
+six:
+    ldx #5
    ; fall through
 load_key_viewport:
    jsr load_viewport
@ -1542,16 +1601,20 @@ copy_byte_loop:
    jsr SETVBV

 main_loop:
-    ; count_frames = 0; count_pixels = 0
+    ; count_frames = 0; count_iters = 0
    lda #0
    sta count_frames
-    sta count_pixels
+    sta count_iters
+    sta count_iters + 1

-    ; total_ms = 0.0; total_pixels = 0.0
-    ldx #total_ms
-    jsr ZF1
-    ldx #total_pixels
-    jsr ZF1
+    ; total_sec = 0.0; total_iters = 0.0
+    jsr ZFR0
+    ldx #.lobyte(total_sec)
+    ldy #.hibyte(total_sec)
+    jsr FST0R
+    ldx #.lobyte(total_iters)
+    ldy #.hibyte(total_iters)
+    jsr FST0R

    jsr clear_screen
    jsr status_bar
@ -1623,38 +1686,32 @@ not_skipped_mask:

 no_key:
    ; check if we should update the counters
-    ;
-    ; count_pixels >= width? update!
-    inc count_pixels
-    lda count_pixels
-    cmp #width
-    bmi update_status

    ; count_frames >= 120? update!
    lda count_frames
    cmp #120 ; >= 2 seconds
-    bmi skipped
+    bpl update_status
+    jmp skipped

 update_status:
-    ; FR0 = (float)count_pixels & clear count_pixels
-    lda count_pixels
-    sta FR0
-    lda #0
-    sta FR0 + 1
-    sta count_pixels
+    ; FR0 = (float)count_iters & clear count_iters
+    copy16 FR0, count_iters
    jsr IFP
+    lda #0
+    sta count_iters
+    sta count_iters + 1

-    ; FR1 = total_pixels
-    ldx #.lobyte(total_pixels)
-    ldy #.hibyte(total_pixels)
+    ; FR1 = total_iters
+    ldx #.lobyte(total_iters)
+    ldy #.hibyte(total_iters)
    jsr FLD1R

    ; FR0 += FR1
    jsr FADD

-    ; total_pixels = FR0
-    ldx #.lobyte(total_pixels)
-    ldy #.hibyte(total_pixels)
+    ; total_iters = FR0
+    ldx #.lobyte(total_iters)
+    ldy #.hibyte(total_iters)
    jsr FST0R


@ -1667,44 +1724,58 @@ update_status:
    sta count_frames
    jsr IFP

-    ; FR0 *= ms_per_frame
-    ldx #.lobyte(ms_per_frame)
-    ldy #.hibyte(ms_per_frame)
+    ; FR0 *= sec_per_frame
+    ldx #.lobyte(sec_per_frame)
+    ldy #.hibyte(sec_per_frame)
    jsr FLD1R
    jsr FMUL

-    ; FR0 += total_ms
-    ldx #total_ms
-    ldy #0
+    ; FR0 += total_sec
+    ldx #.lobyte(total_sec)
+    ldy #.hibyte(total_sec)
    jsr FLD1R
    jsr FADD

-    ; total_ms = FR0
-    ldx #total_ms
-    ldy #0
+    ; total_sec = FR0
+    ldx #.lobyte(total_sec)
+    ldy #.hibyte(total_sec)
    jsr FST0R

-    ; FR0 /= total_pixels
-    ldx #total_pixels
-    ldy #0
+    ; FR0 /= total_iters
+    ldx #.lobyte(total_iters)
+    ldy #.hibyte(total_iters)
    jsr FLD1R
    jsr FDIV

+    ; FR0 *= us_per_sec
+    ldx #.lobyte(us_per_sec)
+    ldy #.hibyte(us_per_sec)
+    jsr FLD1R
+    jsr FMUL
+
+    ; round (down) to integer
+    jsr FPI
+    clc
+    jsr IFP
+
    ; convert to ASCII in INBUFF
    jsr FASC

    ; print the first 6 digits
-    draw_text_indirect speed_start, speed_precision, INBUFF
-    draw_text speed_start + speed_precision, str_speed_len, str_speed
+    draw_text speed_start, str_speed_len, str_speed
+    draw_text_indirect speed_start + str_speed_len, speed_precision, INBUFF

 skipped:

+    ; sx += fill_level[fill_masks] + 1
+    ldx fill_level
+    lda fill_masks,x
    clc
-    lda sx
-    adc #1
+    adc #1 ; will never carry
+    adc sx
    sta sx
-    lda sx + 1
-    adc #0
+    lda #0
+    adc sx + 1
    sta sx + 1

    lda sx
@ -1714,12 +1785,15 @@ skipped:

 loop_sx_done:

+    ; sy += fill_level[fill_masks] + 1
+    ldx fill_level
+    lda fill_masks,x
    clc
-    lda sy
-    adc #1
+    adc #1 ; will never carry
+    adc sy
    sta sy
-    lda sy + 1
-    adc #0
+    lda #0
+    adc sy + 1
    sta sy + 1

    lda sy
--- a/readme.md
+++ b/readme.md
@ -18,7 +18,7 @@ Enjoy! I'll probably work on this off and on for the next few weeks until I've g

 ## Current state

-Basic rendering is functional, with interactive zoom/pan (+/-/arrows) and 4 preset viewports via the number keys.
+Basic rendering is functional, with interactive zoom/pan (+/-/arrows) and 6 preset viewports via the number keys.

 The 16-bit signed integer multiplication takes two 16-bit inputs and emits one 32-bit output in the zero page, using the Atari OS ROM's floating point registers as workspaces. Inputs are clobbered.

@ -27,7 +27,7 @@ The 16-bit signed integer multiplication takes two 16-bit inputs and emits one 3
 * when expanded RAM is available as on 130XE, a 64KB 8-bit multiplication table accelerates the remaining multiplications
 * without expanded RAM, a table of half-squares is used to implement the algorithm from https://everything2.com/title/Fast+6502+multiplication

-The mandelbrot calculations are done using 4.12-precision fixed point numbers with 8.24-precision intermediates. It may be possible to squish this down to 3.13/6.26.
+The mandelbrot calculations are done using 3.13-precision fixed point numbers with 6.26-precision intermediates.

 Iterations are capped at 255.

--- a/todo.md
+++ b/todo.md
@ -1,19 +1,17 @@
 things to try:

-* skip add on the top-byte multiply in sqr8/mul8
-  * should save a few cycles, suggestion by jamey
+* fix status bar to show elapsed time, per-iter time, per-pixel iter count
+
+* 'turbo' mode disabling graphics in full or part

 * patch the entire expanded-ram imul8xe on top of imul8 to avoid the 3-cycle thunk penalty :D

-* try 3.13 fixed point instead of 4.12 for more precision
-  * can we get away without the extra bit?
-  * since exit compare space would be 6.26 i think so
+* maybe clean up the load/layout of the big mul table
+
+* consider alternate lookup tables in the top 16KB under ROM

 * y-axis mirror optimization

-* 'wide pixels' 2x and 4x for a fuller initial image in the tiered rendering
-  * maybe redo tiering to just 4x4, 2x2, 1x1?
-
 * extract viewport for display & re-input via keyboard

 * fujinet screenshot/viewport uploader
Author	SHA1	Message	Date
Brooke Vibber	e0cc704d99	Fix drawing terminator, round usec	2025-01-08 18:34:46 -08:00
Brooke Vibber	7c04862d70	workaround for rounding us/iter for some reason rounding is giving me wrong results not sure what i'm doing wrong :D just show 6 digits :P ok this gets the us/iter working, and it is more stable but the elapsed time still needs to be added	2025-01-05 14:29:27 -08:00
Brooke Vibber	918d15e813	wip us/iter counter seems wrong, gives 32 all the time and that seems too small	2025-01-05 14:05:24 -08:00
Brooke Vibber	eaa00a055a	wip changing time units it does this weird thing where sometimes it's reading out wrong digits and then switches to expected unit of sec/px work in progress no clue what's going on	2025-01-04 18:46:51 -08:00
Brooke Vibber	7e5ca79d9a	move total_ms, total_pixels out of zero page this frees up 12 bytes of zero page space and costs no measurable time as these variables are not in the hot path and there was only a tiny bit different.	2025-01-04 14:25:25 -08:00
Brooke Vibber	d2bf77dc26	todo notes	2025-01-04 12:13:27 -08:00
Brooke Vibber	582ddf497f	apply jamey's suggestion of skipping add for high byte muls rather than saving 0 into the high bytes, then adding the high-byte multiplication later, write it directly in place. this saves a few cycles on every iteration, and it adds up nicely. View 1 overview render times: 130XE: 10.050 ms/px - 4m56s 800XL: 10.906 ms/px - 5m21s	2025-01-04 10:53:51 -08:00
Brooke Vibber	d157fe1306	Faster pixel skipping on 4x4, 2x2 tiers Iterate at fill_masks[fill_level]+1 instead of every pixel and then skipping, saves a smidge of time view 1 with expanded memory: 10.514 ms/px before 10.430 ms/px after	2025-01-04 10:06:12 -08:00
Brooke Vibber	dcf5a3f59e	sixth viewport	2025-01-01 21:15:38 -08:00
Brooke Vibber	837082cf56	tweak viewports skip experimental 6th viewport that got forgotten and limit max zoom to 7 (range 0-7) which is what looks good	2025-01-01 15:45:26 -08:00
Brooke Vibber	65fcb44934	3.13 / 6.26 gives nicer results!	2025-01-01 15:37:12 -08:00
Brooke Vibber	c424f1b8bc	fill in scanlines during tiering	2024-12-31 22:10:27 -08:00
Brooke Vibber	49fe315529	'wide pixels' should get better color on the composite video because the scanlines will be fuller of data	2024-12-31 20:13:11 -08:00
Brooke Vibber	f1ebb21bcb	wip not working wide pixels	2024-12-31 17:49:13 -08:00
Brooke Vibber	87caa52543	add viewport number 5 full zoom	2024-12-31 15:45:03 -08:00